Using basin-scale physiographic attributes to improve river routing in JULES

Land Surface Models (LSMs) simulate various biophysical processes of the terrestrial land surface. They have been developed for a variety of applications, including assessing the impact of modifying a particular process on the ecosystem as a whole, e.g., the impact of climate change on hydrology. Due to their great complexity, developing these models is a continuous and laborious process. For example, the JULES (Joint UK Land Environment Simulator) model is developed by a broad community of inter-disciplinary researchers. However, despite the high level of model development, some processes face parsimonious parameterisation. One of these processes is the routing of surface runoff as simulated by the TRIP (Total Runoff Integrating Pathways) scheme [1]. In its current global parameterisation, TRIP uses uniform velocity and meandering characteristics for the entire land surface regardless of the physiography of the actual river system.

Our work aims to improve the surface runoff's routing by optimising the effective velocity and meandering ratio parameters. In a sample of 360 global river basins, these parameters are correlated with physiographic characteristics to derive a method of extrapolation at the global scale. The development and application of the method were based on river discharge from the global GRDC database [2] and basin-scale physiographic attributes from the HydroATLAS database [3]. A factorial experiment was performed from a combination of 20 setups of effective velocity values and 12 meandering ratios, resulting in a total of 198 simulations. Two optimisation methods were developed; in the first method, the optimum routing parameters are defined for the best NSE improvement with the least deviation from the default routing parameters, whereas in the second method a uniform parameter set was assigned based on a categorisation of the basins. Neural Networks were used for regression and classification, respectively for each method, correlating the optimal routing parameters with the physiographic attributes at the river basin scale. The trained neural networks were applied to the HydroATLAS attributes to extrapolate the routing parameters at the global scale. Simulations of the newly developed river routing configuration showed improved skill in simulating river flow at the global scale (NSE increased by 0.13 on average over 360 global river basins), especially regarding the temporal response. Finally, the present work resulted in a publicly available branch of the JULES code, where spatially varying routing parameters can be introduced, contrary to the globally fixed set.

 

[1] Oki, T., & Sud, Y. C. (1998). Design of Total Runoff Integrating Pathways (TRIP) - A Global River Channel Network. Earth Interactions, 2(1), 1–37. https://doi.org/10.1175/1087-3562(1998)002<0001:DOTRIP>2.3.CO;2

[2] GRDC. (n.d.). The Global Runoff Data Centre, 56068 Koblenz, Germany, 56068 Koblenz, Germany. 56068 Koblenz, Germany.

[3] Linke, S., Lehner, B., Ouellet Dallaire, C., Ariwi, J., Grill, G., Anand, M., Beames, P., Burchard-Levine, V., Maxwell, S., Moidu, H., Tan, F., Thieme, M. (2019). Global hydro-environmental sub-basin and river reach characteristics at high spatial resolution. Scientific Data 6: 283. doi: https://doi.org/10.1038/s41597-019-0300-6

 

Acknowledgement: This work is based upon work from COST Action 19139 - PROCLIAS, supported by COST (European Cooperation in Science and Technology).